🔬 Science & TechMAINS · GS3.13 · GS2.13

India releases AI-in-healthcare strategy SAHI

A recommendatory national framework to guide responsible adoption of AI across India's health system, paired with a benchmarking platform, BODH.

What happened

The Ministry of Health and Family Welfare (MoHFW) released the Strategy for AI in Healthcare for India (SAHI) during the India AI Impact Summit in New Delhi.
SAHI is a recommendatory national framework — a guiding document, not a binding law — for integrating artificial intelligence into India's health system responsibly, anchored in public interest, trust and long-term system resilience.
It is built on 5 core pillars spanning governance and evidence standards; safe, ethical and transparent data infrastructure; and workforce readiness.
A companion platform, BODH (Benchmarking Open Data Platform for Health AI), was launched alongside it to test and validate health-AI tools before they are scaled.
Both are positioned within India's wider digital-health stack, culminating in the Ayushman Bharat Digital Mission (ABDM), and framed as enablers of Viksit Bharat 2047.

Background & context

SAHI does not arrive in isolation; it sits at the head of a policy chain India has been building for nearly a decade. The first anchor is the National Strategy for Artificial Intelligence, released in 2018 by NITI Aayog under the banner #AIforAll, which identified healthcare as one of five priority sectors for AI deployment. SAHI is the sector-specific descendant of that umbrella strategy — where the 2018 document set the national direction across sectors, SAHI translates it into a dedicated playbook for the health system.

In parallel, India laid the digital plumbing that any health-AI strategy depends on. The release traces this trajectory: the National Health Stack (2018, NITI Aayog), the National Digital Health Blueprint (NDHB, 2019), and a National Digital Health Mission strategy overview (2020), all feeding into the Ayushman Bharat Digital Mission (ABDM). ABDM is described in the release as India's first large-scale healthcare digital public infrastructure, with over 860 million health IDs created. This matters because AI is only as good as the data it learns from; ABDM supplies the standardised, identity-linked health records that a national health-AI system needs to function at population scale.

The strategy also reflects a recognised problem the release states plainly. AI tools can already diagnose and predict diseases, streamline clinical workflows, improve hospital management, assist drug discovery and aid research — but adoption in India remains low, and a shortage of diverse, representative data can reduce accuracy and reinforce bias. A model trained on data from one region or demographic can perform poorly, or unsafely, on patients it was never exposed to. SAHI is the government's structured answer to that gap: a way to encourage adoption while guarding against the equity and safety risks that unregulated health AI carries.

Read together, the five pillars trace the full lifecycle of a health-AI tool rather than a single stage of it. Governance and evidence-generation standards set the rules and the proof a tool must meet; the requirement for safe, ethical, robust and transparent digital and data infrastructure covers how patient data is collected, secured and shared; and workforce readiness recognises that clinicians, administrators and technicians must be trained to use, supervise and question these systems, not simply receive their outputs. The strategy's repeated emphasis on diverse and representative data, trust and equity is the thread running through all five — a deliberate guard against the bias problem that has dogged health AI globally, where tools validated on narrow populations have been shown to under-serve the groups already least well served by the health system.

For Prelims

Full form & nature: SAHI = Strategy for AI in Healthcare for India; a recommendatory (advisory, non-binding) national framework, not a statute or regulation.
Released by: Ministry of Health and Family Welfare (MoHFW), at the India AI Impact Summit, New Delhi, on 5 March 2026.
Five core pillars: the framework covers (i) governance and evidence-generation standards; (ii) safe, ethical, robust and transparent digital and data infrastructure; and (iii) workforce readiness — clustered under the five-pillar architecture the strategy defines.
Lineage: builds on the National Strategy for Artificial Intelligence (2018, NITI Aayog, #AIforAll), which named healthcare a priority sector.
BODH: Benchmarking Open Data Platform for Health AI — a structured mechanism to test and validate health-AI solutions before deployment at scale; developed by IIT Kanpur with the National Health Authority (NHA), also launched at the Summit.
Digital-health chain: National Health Stack (2018) → National Digital Health Blueprint (2019) → National Digital Health Mission overview (2020) → Ayushman Bharat Digital Mission (ABDM) with 860 million+ health IDs.
WHO link: at the Summit, the Government of India with the World Health Organization (WHO) published compendia on the Real-World Impact of AI in Health and in Accessibility; WHO's six core principles for AI in health were referenced.
Profiled applications: Scaida BrainCT (AI decision-support for brain CT scans, used in 15,000+ studies across 30+ Tier-2/3 facilities) and SMARTON (voice-first accessibility tool for the visually impaired — 50 languages including 10 Indian languages, 15,000+ users).
Other threads: discussions covered genomics and the "100 million genomes" goal, governance gaps, the need for diverse data, and a stated principle of "Duty of Care."

What it is NOT: SAHI is not a law, Act or binding regulation — it is recommendatory. It is not the National Strategy for AI (that is the 2018 NITI Aayog umbrella document); SAHI is the health-sector framework that builds on it. BODH is the separate testing-and-validation platform (IIT Kanpur + NHA), distinct from SAHI the strategy itself, and distinct from ABDM, which is the underlying digital-health infrastructure. The National Health Authority — not MoHFW directly — partners on BODH and runs the ABDM rails.

The full set (so "how many / match the pairs" survives): SAHI (strategy, MoHFW) · BODH (benchmarking platform, IIT Kanpur + NHA) · National Strategy for AI 2018 (NITI Aayog, #AIforAll) · National Health Stack 2018 · NDHB 2019 · ABDM (860 mn+ health IDs) · WHO six principles for AI in health. Keep the year and the owner attached to each — examiners pair the document with the wrong body.

Why it matters

Health AI is advancing faster than the rules around it, and the cost of getting it wrong is measured in misdiagnoses, not just money. A diagnostic model that works in a metropolitan tertiary hospital but fails in a Tier-3 district facility can widen, rather than close, the access gap it was meant to address. SAHI's significance is that it gives India a single reference point for steering this adoption — setting expectations on evidence, ethics, data security and the readiness of the clinical workforce before tools reach patients at scale.

The pairing with BODH is the operational teeth of an otherwise advisory document. A recommendatory strategy can articulate principles, but BODH provides a concrete gate: an open benchmarking platform where a health-AI tool is tested and validated against shared standards before deployment. That addresses the precise weakness the release flags — low trust and uneven data quality — by making validation a visible, structured step rather than a vendor's claim. Anchoring all of this in ABDM means the strategy is not aspirational plumbing; the data infrastructure for 860 million-plus identity-linked records already exists, giving India a rare combination of population-scale digital health records and a national framework to govern the AI that runs on them.

Set against a peer document, the design choice becomes clearer. The WHO's 2021 guidance on the ethics and governance of AI for health rests on six core principles — protecting autonomy; promoting human well-being, safety and the public interest; ensuring transparency, explainability and intelligibility; fostering responsibility and accountability; ensuring inclusiveness and equity; and promoting AI that is responsive and sustainable. SAHI explicitly references that WHO frame, but where the WHO principles are advisory norms for member states in general, SAHI localises them to India's institutions, its data infrastructure and its workforce, and bolts on a national validation engine in BODH. The result is closer to an implementable system than a statement of values — the same instinct that earlier turned the 2018 #AIforAll strategy from ambition into the ABDM rails now running underneath it.

The two profiled applications make the stakes concrete. Scaida BrainCT, an AI decision-support tool for brain CT scans cited as used in more than 15,000 studies across 30-plus Tier-2 and Tier-3 facilities, is precisely the kind of deployment SAHI is built to govern — a high-stakes diagnostic aid reaching smaller hospitals where specialist radiologists are scarce. SMARTON, a voice-first accessibility tool for the visually impaired supporting 50 languages including 10 Indian languages and serving 15,000-plus users, shows the equity dimension the strategy keeps returning to: AI that widens access for those the system tends to leave out. Both were surfaced in the WHO–India compendia on the real-world impact of AI in health and in accessibility published at the same Summit.

For Mains

Anchor

SAHI can anchor a question on governance of emerging technologies in service delivery — how India is building a regulatory and standards framework for AI in a high-stakes public-service sector (healthcare) before, rather than after, mass adoption.

Data

Use the 860 million-plus ABDM health IDs and the BrainCT/SMARTON deployment figures (15,000+ CT studies; 50 languages, 15,000+ users) as substantiation for the scale and reach of India's digital-health public infrastructure.

Exemplification

SAHI + BODH is a ready example of a "responsible AI" approach: a recommendatory framework backed by a benchmarking-and-validation platform, illustrating how a state can balance innovation with safety, equity and the WHO's six principles for AI in health.

Problematisation

The release itself admits the gaps — low adoption, lack of diverse data, bias risk, governance gaps — which can be cited to problematise the equity and accountability challenges of deploying AI in a diverse, resource-constrained health system.

Way-forward

The "Duty of Care" principle, the genomics/diverse-data push, and the benchmarking-before-scale model offer a way-forward template for any answer on building trust and equity into health technology.

Deploys into: applications of science & technology in everyday life (GS3.13 — IT/AI); health-sector governance and human-resource readiness (GS2.13 — Health/Education/Human resources); and government policy interventions in a sensitive public-service domain.

Ministry of Health and Family Welfare · 2026-03-05 · PRID 2235388 · PIB source ↗